A blog related to original AI artworks. I make AI artworks using coined words and emoji as prompts.

Various Patterns of Fabrics as Negative Prompts

Fabrics come in various patterns. When these are specified as negative prompts, the images generated by Stable Diffusion vary significantly for each pattern.

Argyle

When “Kanabstracd” is used as the positive prompt and “Argyle” as the negative prompt, AI (Stable Diffusion XL) generated the following images, which are based on wavy curves.

The top-left image seems to be influenced by Hokusai, but compared to AI often imitating Hokusai, this one appears to be drawn more conservatively. The waves are colored yellow, a common practice by AI to transition colors from blue or white to yellow. While the color of the waves may seem unnatural, it doesn’t detract from the abstract nature of the painting. The top-right image also appears to depict the sea, with many colorful bubbles drawn. This image predominantly uses blue and black as the colors of the sea, making it more realistic.

The top-left image also resembles calligraphy. A Google image search reveals that Okamoto Tarō created similar works. The top-right image seems to depict calligraphy enclosed in glass. While various glass works with curved shapes can be found through a Google search, none resemble calligraphy. While creating such works in glass would have value, this site only contains images.

Here are some images drawn with slightly looser curves.

The top-left image depicts several spiky black objects, which may resemble viruses or sea urchins. The blue curves might represent the sea. The top-right image depicts a central black object with branching structures, resembling a tree, although the sideways orientation suggests otherwise. There are eye-like objects above, but apart from that, there aren’t many discernible shapes.

For reference, here are examples of images generated with “Argile” as the positive prompt:

These patterns are quite different from the typical argyle pattern, which looks like this:

Damask

When “Kanabstracd” is used as the positive prompt and “Damask” as the negative prompt, AI generated the following images.

The top-left image is quite complex and may resemble a collage. It depicts a seaside scene with a town (perhaps a port town) below. The center also seems to depict a landscape, but it’s difficult to discern. The image below it is simpler, showing a town sloping downward with mountains in the background. There also seems to be something resembling the sea.

For reference, here are examples of images generated with “Damask” as the positive prompt:

Fletching

When “Kanabstracd” is used as the positive prompt and “Fletching” as the negative prompt, AI generated the following lively images.

These images are quite vibrant. The top-left image seems to convey popularity, although it is not depicted clearly. It’s hard to discern what it depicts. The top-right image appears to depict numerous small buildings, but it’s unclear what the swirls around them signify. The sky also seems to contain images of letters, people, and machines.

For reference, here are examples of images generated with “Fletching” as the positive prompt:

These patterns are quite different from the typical fletching pattern, which looks like this:

Marrakech

When “Kanabstracd” is used as the positive prompt and “Marrakech” as the negative prompt, the AI generated the following abstract paintings with distinctive curves. While the curves themselves are similar to those seen in Argyle, the patterns are quite different (perhaps coincidentally).

The first two images are abstract paintings with distinct curves. While the curves themselves are similar to those seen in Argyle, the patterns are quite different (perhaps coincidentally).

The top two images are more linear abstract paintings. The top-left image is almost entirely abstract, while the top-right image depicts something resembling animals or humans. Objects such as legs and buckets can be discerned.

The woman’s face depicted in the top image is quite realistic, surrounded by numerous decorations. The realistic depiction of women is likely due to how Stable Diffusion XL is trained. This image contains a signature, but it’s slightly different from “Kanabstracd.”

For reference, here are examples of images generated with “Marrakech” as the positive prompt:

The original Marrakesh pattern looks like the following, so it is a complete failure.

Check

When specifying “Kanabstracd” as the positive prompt and “Check” as the negative prompt, AI drew the following pictures:

These are landscape paintings drawn with relatively loose curves, each with a different touch. In contrast, the following pictures show a slightly different tendency:

For reference, here are examples of images generated with “Check” as the positive prompt:

These patterns are slightly different from the typical check pattern, which looks like this:

Camouflage

When specifying “Kanabstracd” as the positive prompt and “Camouflage” as the negative prompt, AI drew the following pictures:

Using “Kanabstracd” as the positive prompt often results in the appearance of swirls, which are also present in these pictures. The fourth picture shows swirls in the sky. While this may be a reference to Van Gogh’s “Starry Night,” the influence of Van Gogh is not as evident in this painting. The last four pictures show a different trend.

For reference, here are examples of images generated with “Camouflage” as the positive prompt:

Herringbone

When specifying “Kanabstracd” as the positive prompt and “Herringbone” as the negative prompt, AI drew abstract paintings that evoke various images.

Here, swirls or concentric circles are often depicted. The upper right picture resembles a traffic signal, but only a small part of it can be interpreted as such.

The upper left picture could be interpreted as either a washing machine or a tank, and there are aquatic-like elements visible. The figure of a girl is depicted in the upper left picture, but various other elements are also visible. The upper right picture is more abstract.

For reference, here are examples of images generated with “Herringbone” as the positive prompt:

Showerdot

When specifying “Kanabstracd” as the positive prompt and “Showerdot” as the negative prompt, AI drew the following pictures:

These pictures are drawn with a unique style. Although specifying “showerdot” may result in different pictures, those listed above are representative.

For reference, here are examples of images generated with “Showerdot” as the positive prompt.

The image in the top left corner is the closest to resembling shower dots. As for the image in the top right corner, while it could be said to depict shower dots, it also includes human figures. Additionally, here are a few more images generated with “Showerdot” as the positive prompt.

The image in the top left corner no longer resembles shower dots, but since it includes numerous shapes resembling circles, it might have a somewhat similar effect when specified as a negative prompt. The image in the top right corner appears to depict rain along with faces. What is being depicted is certainly water. However, although there are some shower dots, they are few, so it cannot be said to truly represent the image of shower dots.

Scrollwork

When specifying “Kanabstracd” as the positive prompt and “Scrollwork” as the negative prompt, AI drew the following pictures:

For reference, here are examples of images generated with “Scrollwork” as the positive prompt:

Tartan

When specifying “Kanabstracd” as the positive prompt and “Tartan” as the negative prompt, AI drew relatively simple pictures as follows:

The above two pictures feature intricate wavy lines.

These also include swirls or concentric circles. The bottom three pictures have quite different trends.

For reference, here are examples of images generated with “Tartan” as the positive prompt:

The following image is a tartan pattern obtained from the internet, and it can be said that the upper image is quite similar to this.

Checkered

When specifying “Kanabstracd” as the positive prompt and “Checkered” as the negative prompt, AI drew relatively simple pictures as follows:

The above two pictures predominantly feature linear shapes, with intricate curves also drawn simultaneously.

The above three pictures feature “parallel” curves. However, especially in the third picture, there are other elements as well. The last two pictures show different trends.

For reference, here are examples of images generated with “Checkered” as the positive prompt:

Fairisle

When specifying “Kanabstracd” as the positive prompt and “Fairisle” as the negative prompt, AI drew the following pictures:

For reference, here are examples of images generated with “Fairisle” as the positive prompt:

Gingham

When specifying “Kanabstracd” as the positive prompt and “Gingham” as the negative prompt, AI drew the following pictures:

For reference, here are examples of images generated with “Gingham” as the positive prompt:

Houndstooth

When specifying “Kanabstracd” as the positive prompt and “Houndstooth” as the negative prompt, AI drew the following pictures:

For reference, here are examples of images generated with “Houndstooth” as the positive prompt:

Plaid

When given the positive prompt “Kanabstracd” and the negative prompt “Plaid,” the AI produced the following images.

AI often draws images resembling Van Gogh’s “Starry Night” or Munch’s “The Scream,” but the top-left image appears to combine elements of both. When searched on Google Images, both of these images can be found.

For reference, here are examples of images generated with “Plaid” as the positive prompt: